#27 October 30, 2018

Evolution of the Kubernetes Community, with Sarah Novotny

Hosts: Craig Box, Adam Glick

Sarah Novotny is Head of Open Source Strategy at Google Cloud and a board member of the Linux Foundation (the parent of the CNCF). She joins Craig and Adam to talk about the evolution of the Kubernetes community, governance models and Codes of Conduct, and how nascent open source communities can learn from it.

Do you have something cool to share? Some questions? Let us know:

Chatter of the week

News of the week

ADAM GLICK: Hi and welcome to the Kubernetes Podcast from Google. I'm Adam Glick.

CRAIG BOX: And I'm Craig Box.

[MUSIC PLAYING]

ADAM GLICK: I wanted to start this week by saying, happy birthday, Craig.

CRAIG BOX: Oh, thank you, Adam. Where did that come from? That's so kind of you.

ADAM GLICK: I just wanted to recognize that. We seem to have touched on a couple of birthdays recently, so I wanted to congratulate you on yours.

CRAIG BOX: It's not so much about that. It's about me telling you what I did for the weekend, was which lovely birthday-related activities.

ADAM GLICK: Did it involve cake?

CRAIG BOX: It did. There's still half a cake. If you want some, it's just downstairs. Feel free to pop by. [LAUGHS]

ADAM GLICK: I'll get on a plane post haste.

CRAIG BOX: So Fern arranged a very lovely day for me, where I got to do a tour of the Lord's Cricket Ground. How good's your knowledge of cricket grounds?

ADAM GLICK: Poor.

CRAIG BOX: Well, they're oval shaped. Did you know that?

ADAM GLICK: I did.

CRAIG BOX: Well, there you go. So my friend Sam and I did a tour of the hallowed turf of Lords. And then we went and saw a play in the West End called, "The Play That Goes Wrong," which I believe is on Broadway as well. And it does tour around the place, but it's sort of a slapstick comedy in the Monty Python vein. It was very funny.

ADAM GLICK: Nice.

CRAIG BOX: And then topped it off the day afterwards by seeing the new Queen film, "Bohemian Rhapsody."

ADAM GLICK: That looked good. Would you recommend it?

CRAIG BOX: I've seen reviews that basically describe it as a two-hour-long Wikipedia article, which I think is kind of true in a way. So as a fan of the band, someone sort of familiar with the story, it sort of hit all the beats. And a lot of people look at it and say, oh, those things didn't happen in that order, or so on. But the lead actor, Rami Malek from "Mr. Robot," he did a very good job. He's tipped for a possible Oscar nomination for inhabiting the teeth and the role of Freddy Mercury.

ADAM GLICK: Speaking of which, if you haven't seen "Mr. Robot," I watched season one and really appreciated both what they do with the show and some of the meta things they do around show names and things in there to have a whole bunch of Easter eggs. And if you get into that kind of thing, drilling into the various kind of subplots and things they have going on online, there's a wonderful rabbit hole you can go down of the various puzzles, and hints, and websites that you'll stumble across. If you love that kind of thing, and I do, you'll get a kick out of it.

CRAIG BOX: It's one of those shows that you absolutely have to not have the secrets spoiled halfway through the first season. If you accidentally read something, it's like, oh, I would have much more enjoyed watching it not knowing that.

ADAM GLICK: Oh, yeah, definitely.

CRAIG BOX: Try not to know anything going in.

ADAM GLICK: If you haven't, don't read about it first because something will definitely spoil the outcome. Speaking of geek media, season two of "Castlevania" is out, which I started watching. And for those of you who were fans of the Nintendo game series, this is definitely a more mature cartoon.

CRAIG BOX: See I was confused when you mentioned this because my recollection is, hmm, "Castlevania," wasn't the a Nintendo game?

ADAM GLICK: It was. And it is--

CRAIG BOX: What's it doing on Netflix?

ADAM GLICK: They're in season two of an animated version. And if you are interested in-- if someone were to build out the back story of that world, that's a good place to go. I've really been enjoying it.

CRAIG BOX: Will the "Legend of Zelda" be next?

ADAM GLICK: You'll have to wait and see.

CRAIG BOX: Well, while we wait, let's get to the news.

[MUSIC PLAYING]

ADAM GLICK: The news of the week is, of course, IBM's acquisition of Red Hat. The $34 billion acquisition is being billed as the largest software acquisition ever, dwarfing Microsoft's $7 billion acquisition of GitHub, which closed this week. It represents a 60% premium over Red Hat's closing stock price. In after-hours trading, IBM shares were down 5% and Red Hat's up almost 50%.

Red Hat will remain an independent division within IBM's hybrid cloud, with Red Hat's CEO joining IBM's senior management team and reporting directly to IBM's CEO. IBM says they intend to maintain Red Had's headquarters, facilities, brand, and practices, and that they see this acquisition as giving them a leadership position in the hybrid cloud market. Obviously, Kubernetes is a big part of Red Hat's recent success.

CRAIG BOX: Remember when we told you we couldn't wait to see what was on tap in Red Hat's OpenShift 3.11 for workgroups? The answer is "more", with additions based on the CoreOS acquisition, and "less", with a new Container Engine offering.

In the "more" category, tools first shown in CoreOS's Tectonic platform have moved to OpenShift, including a console for ops teams and Kubernetes operators for applications like Couchbase, etcd, and MongoDB. Who watches the watchers? There's a tech preview of an operator lifecycle manager, otherwise known as the Operator operator.

Red Hat has also acknowledged that a large number of their customers and prospects want just a supported Kubernetes platform without having to pay for developer tooling. They have released their Container Engine offering, which provides Kubernetes support at a lower price point. Features such as their source to deployment capabilities and their upcoming SDO-based service mesh will remain differentiators for the full OpenShift Container Platform.

ADAM GLICK: IBM has also recently announced that their cloud Kubernetes service is now available in one presumably very fashionably-dressed zone in Milan, Italy.

CRAIG BOX: Mirantis, a company previously known for their open stack distribution has released Mirantis cloud platform, Edge, a new lower footprint Kubernetes-based effort to run at the edge of the network with a target size of six node deployments. Customers will be able to run both containers and VMs on the MCP Edge using the Virtlet technology for running VMs on Kubernetes.

ADAM GLICK: Mesosphere announced general availability of Mesosphere Kubernetes Engine, which targets people running multiple Kubernetes clusters across their company. This offering has the ability to manage multiple clusters from one control plane and the ability to have multiple clusters share machines using their DC/OS scheduler underneath.

CRAIG BOX: It sounds like it was a marathon effort.

Steven Acreman from the Kubedex has decided to take on the evaluator-in-chief role in the Kubernetes community and has this week added a post comparing some on-prem distributions. While he is but one person and has only chosen to evaluate certain tools, his source spreadsheets are open for editing. And we encourage anyone who wants to detail their favorite vendor to do so. One of Steven's recent tools is called Dolos, the Greek word for the company that runs the theme parks in Westworld.

Dolos measures the time to turn up a cluster and turn it back down again. Initial results are in for Google Cloud, Azure, and Amazon. And while it may be a reasonably uncommon use case to turn clusters up and down, Steven says that a lot of people do this with their testing environments and believes it to be a worthwhile point of comparison between the vendors.

ADAM GLICK: gRPC-Web is going GA. gRPC-Web is a JavaScript client library that enables web apps to communicate directly with back end gRPC services without requiring an HTTP server to act as an intermediary. This means you can have end-to-end RPC communications using protocol buffers without creating any JSON or having to do any serialization.

CRAIG BOX: Andrew Chen and Dominik Tornow continue their series of articles applying systems modeling to the underlying concepts of Kubernetes. This week, they provide a deep dive into exactly what a pod is and does, explaining with words, diagrams, and what my vague memory of university wants to call sit theory.

ADAM GLICK: A FoundationDB summit has been announced as part of KubeCon Seattle this December. FoundationDB is a distributed ACID compliant data store designed from the ground up to be deployed on clusters of commodity hardware. The company behind it, also named FoundationDB, was acquired by Apple in 2015, and the software was unexpectedly open sourced this April. This will be the inaugural FoundationDB Community Conference.

CRAIG BOX: If you can't make it to one of the CNCF's flagship events in North America, Europe, or China, you will be pleased to know that they are looking to expand into other parts of the world with a series of single track, single day events in 2019. Dan Kohn, the head of the CNCF, said in an interview he hopes to hit Bangalore, Sao Paulo, Johannesburg, and possibly Malaysia, targeting areas with a lot of software developers who don't necessarily have the resources to travel. We'll keep you up-to-date as planning progresses.

ADAM GLICK: And that's the news.

[MUSIC PLAYING]

CRAIG BOX: Sarah Novotny is the head of Open Source Strategy for Google Cloud. She sits on the Board of Directors for the Linux Foundation, the Node.js Foundation, and as Google's alternate board member for the Cloud Native Computing Foundation. She is additionally, a CNCF ambassador and is program chair emeritus for O'Reilly Media's OSCON. Welcome to the show, Sarah.

SARAH NOVOTNY: Thanks. It's great to be here.

ADAM GLICK: Sarah, how'd you get started in open source?

SARAH NOVOTNY: I started, as so many people do in open source, as a consumer. My first job in technology involved me being handed a stack of floppy drives and being told to call my boss when I had my Red Hat Linux box on the network. At the time it required a kernel recompile and new network drivers being installed. So it was a long time ago and as a user.

From there, I went on to take that user perspective and work in the MySQL community as an advocate for MySQL working at great length, giving talks about different use cases that my customers had and learning very much the developer relations strategy inside of the house with open source projects.

ADAM GLICK: And when did you start working with Kubernetes?

SARAH NOVOTNY: Kubernetes was actually super fun. Kubernetes came to me in some ways through OSCON. So I worked, as Craig mentioned, as program chair of OSCON for many years. And that involved making sure that there was a good solid editorial story about all of the different open source work that was happening and bringing to OSCON, as much as possible, the different new and interesting technology. So I got to meet Craig McLuckie who wanted to make a big announcement at OSCON in 2015 about Kubernetes, and it had just reached 1.0, and there was going to be a foundation created around this.

And so for about six months before that, I was learning all about what Kubernetes was doing, and all about how Google wanted Kubernetes to really change the industry, the big strategy behind Kubernetes as well as how the intent was to take this out as an open source project, and then how we had to really invest in the community in order to enable the strategy of changing the way the industry spoke about portable workload units in the cloud from virtual machines to containers.

CRAIG BOX: Were you working at Google by this point?

SARAH NOVOTNY: I was not working at Google at this point. These conversations actually--

CRAIG BOX: That's a lot of things for him to ask before you've even started.

SARAH NOVOTNY: Well, it was fascinating because we had a lot of long conversations about what would the community look like, what could the community look like? And it was David Aronchick actually who was like, why don't we just bring her in and have her help us do this? So that started a series of conversations that ultimately turned into a question of do you want to come into Google to do this?

And oh, by the way, can you help us do this Cloud Native Computing Foundation thing, and what would that look like, and where would you put it, and how would it be structured? And lots of fun back then. I mean, it's still lots of fun, but lots of potential fun in that space as opposed to ongoing chaotic fun now.

CRAIG BOX: Stepping back a bit and looking at open source communities around that time and your involvement in communities like NGINX and MySQL, for example, how do you think overall open source moved from being a technical meritocracy to a more community-driven situation where foundations exist and with running things in more open fashion?

SARAH NOVOTNY: It's interesting you use the phrase technical meritocracy, because that has certainly fallen out of favor at this point as a way to look at how open source projects are structured. Because there's so much more now to an open source project than strictly code contributions. So there needed to be an evolution from thinking about open source as just who has the clearest vision of the technical architecture, and who has been with the project longest, and who has the greatest skill in reviews and familiarity with the nuances and quirks of a particular codebase to a much more holistic, much more inclusive project concept that looks more like a company and less like an engineering division.

And a lot of this has changed as open source projects have moved from being more exclusively infrastructure projects. So engineers writing for engineers, to engineers writing for operators, engineers writing for end users. Actually two great examples of projects which have done a very good job of writing open source software for end users include WordPress and Drupal.

The whole idea is that this is not engineers for engineers. And so there is a much higher standard of working together as a whole toward building a product as opposed to writing code which executes properly and is perhaps the most elegant code possible. So I think that that is an ongoing process that has happened.

And actually I was at a conference just this week, Sustain Open Source or Sustain OSS, which spent a lot of time talking about how to make it clear as a value of projects that accessibility, user interface, usability, documentation, and general product vision are all just as important and need to be rewarded from contributors just as much as actual pull requests and code commits.

ADAM GLICK: What made you decide to pick the Kubernetes community to take a leadership role in?

SARAH NOVOTNY: I think everyone has people that they watch in their communities for a signal about a new thing that they expect to be a really big change and have an opportunity to fundamentally alter how our industry works. And in 2014 and '15, I was starting to hear noises from those lighthouses in my community about, there's this thing, and it's super cool. And you know how we all loved zones in Solaris, and it was really useful to have this whole process isolated concept of containers.

And you know how it was never quite working? Well, it was never quite working easily. It worked very well, but it didn't work easily. And it was becoming more and more easy to work with containers. And then people were starting to play with containers.

And then as people were playing with Docker, as it started to grow, there became this inflection point where people were like, this is great and now I have 1,000 of them or 10,000 of them. And knowing the history of Google and Google's container work, I knew that there was a lot of experience inside Google that we brought to this container concept and how to rationalize and even reason about as humans where containers are, how many we need when we have them, why would we use the meat cloud to restart them when we could use automation?

And so as I was hearing Kubernetes coming as a concept being talked about, started with Google, started with Red Hat, started with other partners-- CoreOS being another-- I knew that this was going to be a very interesting project. And so I started paying a lot of attention to it.

ADAM GLICK: I think that's the first time I've ever heard the term "meat cloud" used.

SARAH NOVOTNY: Really? Oh, meat cloud is fantastic. You do know what I mean though, yes?

ADAM GLICK: I know exactly what you mean. I used to think of things like sneakernet, snail mail, so meat cloud apparently is the latest generation of cloud provided, human powered automation.

SARAH NOVOTNY: Exactly. Yes, the meat cloud. Why would we use the meat cloud to restart containers? That seems super silly.

ADAM GLICK: So what are some of the biggest changes that you've seen in the Kubernetes community given how long you've been a part of it?

SARAH NOVOTNY: So when I give the very shorthand answer to this, people ask me, what was I hired to do at Google? And I say I was hired to take Kubernetes from a company-led project to a community-led project. And that's a pretty fundamental shift, so I'd say that that's one of the biggest changes.

And then absolutely everything else which are many, many large changes feed that. So some examples would be early on, so 2015, '16 again, there was-- I refer to it as the shadowy cabal of leadership-- which is sort of if you ask one of the people who calls themselves a founder of Kubernetes then they might be able to find you the right person to say yes or no to a thing. But if you don't like that answer, you could go ask a different one.

So there wasn't really a clear governance model at that point. And that led to a lot of uncertainty and disempowerment of the community. And so making sure that we built in a transparent governance and even evolving that transparent governance was a huge change. Because how do you start building a model to manage a group of people who don't, by default, have a way to define themselves as a specific narrow group.

Like who is a contributor was an open question at that point. Who is a collaborator? Who is someone who is just interested in the community? So we didn't have a group that could even choose a leader much less decide how or what leadership structure we wanted. So working our way through governance in that sense and figuring out from the broad group of the Kubernetes community how we wanted to govern was a huge, huge change.

There was a great joke that came out in one of the KubeCon events that showed up as a tweet, and I wish I could find who said it. And maybe if somebody on the podcast remembers who said it or who tweeted it, they can send it to me. But the tweet after KubeCon was like, "wow, amazing KubeCon, blah, blah, blah. I think I met all seven of the four founders of Kubernetes".

ADAM GLICK: [LAUGHS]

CRAIG BOX: It's been gratifying to see Kubernetes succeed as a community. But obviously, there are alternatives in the space and any one of them could have got the impetus behind it. If you are starting off an open source project, and you're in that situation where you hope you succeed, but you don't know if you will, when is the right time to bring in that formal governance? Is it too soon to do it when you need to focus on growing your success and making sure that you do get to grow out of that embryonic phase?

SARAH NOVOTNY: Formal governance is always living. And every group going through every life cycle stage of a project will have slightly different needs of governance. And what I've seen in Kubernetes and worked in other communities trying to help develop is, at the stage you are, you want to have the minimum level of governance that allows you to alleviate discomfort, decrease the amount of FUD and increase the amount of trust in the project and the leadership while always having a clear path to change that governance.

So you really are building the plane, the train, you're building the project as you are using it. And so there needs to be a way to evolve it that is agreed upon. Even if that agreed upon way to evolve it then says, yes, what we're doing now is awful. We have to do it completely over.

You have a method, or you have a starting point, or you having group who says or can say, we need to start over, or we need to change this, and that group is accountable. And that might include a group or a one line definition of in the event of no confidence, the following happens.

CRAIG BOX: If you're an engineer starting a project, and it's not your strong suit dealing with people, for example, how can you help get that formalized?

SARAH NOVOTNY: I think there will always be people in your community that you trust and that you look to for help in those situations. So even if this isn't your strong suit, you probably know someone in your community who is more skilled at that. And having that trust and then using that trust to establish commutative trust and commutative leadership is super important. I am a firm believer that one leader is a very difficult way to run any project, and it is a great risk to any project.

So making sure that there are lieutenants, that there are always leaders being grown, that there's always some sort of succession planning built into this. Even if it's the most limited like the only other person who can do this is so and so. So if I can't, please use them. But making sure that there's always a growth path and an escape hatch for losing a single person or losing a group of people within a project is just good management of any group.

ADAM GLICK: How much of this organizational design was effectively inspired by different governmental systems? This reminds me very much of people setting up a new governing system from a political standpoint. And how do you answer the questions of benevolent dictatorship versus messy parliamentary system, succession planning. That sounds a lot like sitting down and having a constitutional congress, so to speak.

CRAIG BOX: Is this the worst possible system excepting all the other ones?

SARAH NOVOTNY: Exactly. It is the worst possible system except for all the rest. So there are definitely markers of this. There are a few key differences. One, goodness, I certainly hope we will never have life and death decisions in an open source project, so hooray for that. So that's a huge exception in it.

The second one. We often look to politics or political systems for examples. In setting up the Cloud Native Computing Foundation, for example, the by laws built out three leadership groups. Very much an echo of the US system of three leadership groups. So we have the End User Council, which is supposed to be representing, obviously, by name, the end users.

We have the Technical Oversight Committee who is leading through industry technological experts. They may not be directly working on any of the specific projects, but they have that big holistic view of what does Cloud Native mean. And then there is the governing board, which is the business interests and vendor interests in a lot of cases of the Cloud Native ecosystem as well.

And these three bodies are supposed to hold each other in check. So there's definitely reference to a lot of these governmental systems. And as I was building, or as we were building-- because it really was very community built, the Kubernetes' governance system-- I was reading a really good book about continuity of government called "Raven Rock."

So I recommend that book to everyone. It's a really interesting book about the complexities of what if your leadership is cut off at the head. How do you reestablish any sort of order? And these are all things that you have to look at in large group management.

Now, your average small open source project-- 10, 15, 100 contributors-- is not going to need the same level of process and policy that you might in a larger project. But for Kubernetes, as an example, we were trying to build a distributed compute orchestration system, so a platform for distributed computing. And this suggested, if we go back to looking at Conway's law, that we really needed a distributed decision making system.

And there have been good and bad pieces of that, but each of the special interest groups inside Kubernetes owns a portion of the code and has decision making authority and ownership and some amount of architectural authority over those subcomponents. Where those subcomponents interact with other subcomponents that have different owners, that's where our overarching architectural group's architecture gets involved. Or they get involved if there are disputes between two groups.

Ideally, two groups that have meeting subsystems come to a conclusion on their own. And it's the right one, and off they go, and it never comes to the architectural group or to the steering committee in Kubernetes. But if there is a dispute, and it can't be worked out, then it goes to the architectural committee for a technical answer. Redundancy. It's all about redundancy.

CRAIG BOX: I was gonna make a Paxos generals joke about the number of management groups.

SARAH NOVOTNY: Feel free to. The ideal, though, is that we push the decisions as close as we can to the people writing the code. It's really only during disagreements that we would ever go up any sort of management-- ladder is not quite the right way even to look at it-- that we would go seeking further opinions.

CRAIG BOX: There have been a number of high profile debates about codes of conduct in committees lately. You've been through the process of implementing a code of conduct for the Kubernetes community and then effectively upstreaming that to be something that's now CNCF Project. Did you see some of the same problems in creating that code of conduct? And what lessons do you think that you would like to impart to people who are doing similar things today?

SARAH NOVOTNY: The question of the week, the month, the year. I was very lucky in that this community has a lot of very experienced and established project leaders who have worked at large companies and understand the value of cultural norms. So even before I started with Kubernetes, there was a code of conduct based on one of the industry standards that was in place in the project.

I had the luck of being the person who enforced it for 2, 2 and 1/2 years. And even better luck in that our community was very well empowered to do peer encouragement around good behaviors, and setting good standards, and setting good cultures. So there were very few things that bubbled up to actual code of conduct incidents that had to be investigated or managed.

I think it is a super important portion of all open source that there are community expectations and that those expectations are clearly defined. We're moving to a space in this conversation broadly across the open source world where there is also a clearly defined set of outcomes if those expectations are not met. So there is an enforcement portion, so that you have clear expectations of if someone is in breach of the code of conduct in a particular type of way, this is how they can expect that the community will respond.

Mozilla actually is doing a phenomenal job with this work. They've got an open working group about code of conduct. They are doing a lot of work trying to make sure that they codify in a lot of ways what violations require what type of response and then how someone might expect to have an interaction with the code of conduct committee on a multiple violation or something like that.

So really what we found-- I found-- is the more you empower your community to hold good standards and the more your leadership embodies those standards, the better the experience will be and the less you will need the enforcement. That said, there should always be the clear model. This goes back to transparency and open source transparency of governance, transparency of culture, transparency of values, expectations of engagement, and now enforcement in the event that someone needs to be moderated or mediated or any sort of work in that sense needs to be intervened.

CRAIG BOX: How should we grow contributors today to ensure they don't have similar empathy problems that have been on display with past leaders in open source communities?

SARAH NOVOTNY: I think this goes back to that peer empowerment and peer engagement. Making sure that everyone who is working in a community is empowered to engage with and encourage the cultural norms of the project. And those cultural norms are set very much by the leadership.

The values, I mentioned those. I think she's going to get tired of hearing that I keep referencing it, but there's a great article that Aja Hammerly wrote, which I think is entitled "We Don't Do That Here." And it's a really clean jujitsu move to take something that is uncomfortable and not police someone. But just let them know not that's not cool.

We don't call ideas stupid here. We just can't do that here. We say, your idea has some flaws that look like this. Or have you considered these edge cases, because your model breaks down. It might work better like this.

So a very soft but firm empowerment to guide a community in a way that shows that A, we care, B, there are consequences, and C, those consequences don't have to be dire. They can just be a conversation of, hey, you're new around here. We should do this slightly differently here.

ADAM GLICK: When you look at the Kubernetes community it's been growing so quickly. So many more people are getting involved. It's incredibly vibrant, which makes all of us very excited. Where do you think it's heading next? What are the next challenges that you think that the community will tackle as it grows and becomes more and more a part of the core software stack people use?

SARAH NOVOTNY: I think there are a couple, and one of them is going to be funny, so I'll start with that one. And one of them's going to be more serious and longer.

CRAIG BOX: Is it too many podcasts?

SARAH NOVOTNY: It is not yet too many podcasts.

[LAUGHING]

OK. I think actually, though, one of the things the community is going to struggle with is like all infrastructure, this should be transparent and boring. It will become transparent and boring. And then it's not going to be nearly the hype that it has been for the last several years.

And I think that's going to be a really interesting evolution for everyone in this space. It's trying to go from being the coolest kid on the block-- and what nerd does not want to be the coolest kid on the block even in a tech group-- to the space where this is suddenly transparent and boring. So I think that's one challenge.

I think the other challenge is actually one where we have to manage the growth that we're seeing in this. So I gave a talk last year at KubeCon which I thought had just blown out the numbers as to what a KubeCon could possibly look like in Austin. And it was like 3,800 people, if I remember correctly. I'm pulling that totally off the top of my head. I may be off by a few hundred.

But I was like, we have to be really intentional about our values and is this success if we have a KubeCon next year that is 8,000 people? And is this what we want from the community? Is this the way we want to grow? Is this the right choice, and can we uphold the values of the community in that sort of growth?

And I was more prescient than I ever intended, 'cause KubeCon this year I think is going to cap out at about 7,000 people, this KubeCon in Seattle. And that is because of the venue that we are actually going to have to close down tickets not because there isn't interest anymore.

So I think the growth of the community, and the growth of the values, and the growth of civil discourse, and the growth of leadership in this community is the piece that we are going to be struggling with most over the next, or have been struggling most with even in last year and will continue to. It's like what is the definition of success here? Is it just more hype?

I hope not. That would not, to me, seem like a great definition of success. Or is it that it becomes boring infrastructure and the numbers decrease at KubeCon? If that's the case, then that's kind of counter-intuitive, but it might be a success case.

So defining really what we think of as success here is going to be one of the next big opportunities. And then trying to manage the growth, either positive or negative growth, to get there.

CRAIG BOX: All right two final questions for you. People who have seen your picture on conference websites or received email from you may have seen you hiding behind a Viewmaster or a giant gold heart. What's up with that?

SARAH NOVOTNY: [LAUGHS] So for both of those, I have to thank Julian Cash. He's a photographer in San Francisco who has attended many open source events and has set up these just amazing fun spaces where he's got a pile of toys, and he takes photos on white backgrounds. And so you'll see lots and lots of technologists have these sorts of photos.

For me, picking the Viewmaster was really about changing my perspective and telling stories in combination and making sure that I'm telling stories in a way that are accessible to people who may not have the same perspective as I do. So when I picked that one up and I was playing with it, that was what was going through my head.

And then the heart of gold, I actually brought to Julian Cash. So I brought my own toy to his photo shoot that year. And at the heart of gold, for me, was both an homage to the "Hitchhiker's Guide to the Galaxy," of course, but also the fact that communities in technology often forget the human aspect of their engagements and often over weight toward the technical aspects.

But you can really grow your community by paying attention to people as whole humans and get that much more technical excellence out of them, or community excellence out of them, or design excellence out of them by engaging with them as people. So many thanks to Julian Cash on those.

ADAM GLICK: Awesome. It's been great chatting with you, Sarah. If people want to catch up with you either in person or online, where can they find you?

SARAH NOVOTNY: In person, I am based in the San Francisco Bay Area and up in Seattle a fair amount. And the last two events that I hope I get to go to in this year are KubeCon China and then KubeCon North America. So I will be attending both of those in the next two months.

And then online, you can find me pretty much anywhere as SarahNovotny in a single word. That includes Twitter. I am spending less and less time on Facebook, but you can still find me there. And LinkedIn is the same.

CRAIG BOX: Sarah, thank you so much for joining us today.

SARAH NOVOTNY: Happy to. It's been great fun chatting with you all.

[MUSIC PLAYING]

ADAM GLICK: Thanks for listening. As always, if you enjoyed this show, please help us spread the word and tell a friend. If you have any feedback for us, you can find us on Twitter @kubernetespod or reach us by email kubernetespodcast@google.com.

CRAIG BOX: You can also find our show notes at our website at kubernetespodcast.com where you'll find transcriptions of each episode a few days after they air. Until next time, take care.

ADAM GLICK: Catch you next week.

[MUSIC PLAYING]